The present invention relates to an image processing apparatus and method and, more particularly, to an image processing apparatus and method which recognize the three dimensional geometric shape of a stereoscopic image using the right and left images of the object.
For recognizing a three dimensional geometrical shape of an object using right and left images of a stereoscopic image picked-up by a stereoscopic camera or the like, there is a conventional system for finding three dimensional coordinates by finding all corresponding points in the right and left images by a matching process and the like.
However, in the above-type system, since a computer automatically finds the corresponding points, there is a fear that the located corresponding points may not actually correspond to each other. Further, because all the corresponding points are searched, the computer has to perform great many calculations and it requires considerable time to obtain the three dimensional coordinates.
For the purpose of overcoming these problems, the assignee of the present invention has submitted a patent application regarding a method where a user interrupts the computer and designates corresponding points for finding a three dimensional shape in Japanese Patent Application No. 5-299515 (applied on Nov. 30, 1993). According to this method, after the three dimensional coordinates are found from corresponding points of the right and left images of a stereoscopic image, a triangle connecting the three dimensional coordinate points is generated, thus forming a plane. In the method proposed by the assignee, a triangulation net method and the like based on the Delaunay's principle is used for generating the triangle.
Generally, a plurality of three dimensional objects are often included in an image. If there are a plurality of objects in an image, all the objects including background are always dealt with as a single three dimensional shape when a conventional method is employed. In another case, it is possible to deal with only an area portion, designated by a user interruptedly, in the image as a single object. More specifically, even if a plurality of objects are included in an image, there is no apparatus capable of recognizing the objects separately in three dimensional space. Therefore, the plurality of objects are always dealt with as a single object (including objects constructing background).
Further, right and left images are only used for inputting three dimensional shape, and have not been used as texture information.
Furthermore, there is a problem in which, when a method for generating a plane uses Delaunay triangulation net, if there is a three dimensional concave area in an object, the concave area can not be recognized.
Further, it is necessary to designate all the corresponding points in the right and left images manually, a user has to perform quite a lot of manual operations. In addition, it is difficult to designate corresponding points in the right and left images of the stereoscopic image precisely by manual operation. Especially, if there is no clear texture information in the image, to designate the corresponding points is even more difficult.